The present invention relates to a library apparatus using a magnetic tape cartridge, an optical disk cartridge, or the like as a memory medium and, more particularly, to a library apparatus which receives a plurality of medium conveyance information and can simultaneously drive two medium conveying apparatuses.
In a recent computer system, in association with an increase in number of media which are used in an MTU apparatus, an optical disk apparatus, and the like as external memory devices, a high efficiency of the operation, easiness of the maintenance, a reliability of the apparatus, or the like is required. For this purpose, a library apparatus which can perform an automatic operation such as storage of the media and the conveyance to the apparatus is provided. When the number of operations of the Jobs or the number of connections of a host computer increases, however, since medium conveyance information increases, it is necessary to efficiently execute them.
FIG. 1 shows an outline of a conventional library apparatus. A library apparatus 400 is used as an information stocker of upper-order host computers 20-1 to 20-4. For example, a director 402-1 has four channels (A to D) and a director 402-2 has four channels (E to H). In this case, the channels (A and B) and the channels (E and F) are used. A cell block 410 as a medium stocker to enclose media such as magnetic tape cartridges or the like is provided in the library apparatus 400. Two accessors 408-1 and 408-2 as medium conveying apparatuses are provided on a rail 414 along the cell block 410. The accessor 408-1 is directly controlled by a machine controller 406-1. The accessor 408-2 is directly controlled by a machine controller 406-2. An accessor controller 404 is provided as an upper-order apparatus for the machine controllers 406-1 and 406-2. Generally, the accessor 408-1 operates and executes the conveyance control and the accessor 408-2 doesn't operate. This is because since two accessors 408-1 and 408-2 operate on one rail 414, they cannot simultaneously operate and they are unitarily managed by one control section so that they individually operate by an exclusive control method. The accessor controller 404 exists as a spare apparatus for alternation when the accessor 408-1 causes a fault or the like and enters an unexecutable state.
FIG. 2 is a diagram showing the operation when a moving command (a medium conveying command) is received in the conventional apparatus. Physical machine number addresses of the directors 402-1 and 402-2 are defined as D0 and D1 and physical machine number addresses of the accessors 408-1 and 408-2 are defined as AS0 and AS1. It is now assumed that the host computer 20-1 generated a moving command from the channel (B) to a logical machine number address (logical input/output device address) #0. A logical machine number address #8 is used in an actual apparatus. When the moving command is received from the channel (B), the director 402-1 instructs the medium conveyance to the accessor controller 404. The accessor controller 404 receives the conveying command from the director 402-1 and refers to an executing state management table 422. In this instance, since an execution flag is not set, it is confirmed that the accessors 408-1 and 408-2 can be used, and a response indicative of the reception of the conveying command is sent to the director 402. The director 402-1 subsequently confirms the reception response from the accessor controller 404 and sets the execution flag at the position of the channel (B) in a machine No. table 418-1 and reports an initial status to the channel B. After that, the director 402-1 receives the medium conveyance information `XXXX` as a command parameter from the channel (B) and transfers to the accessor controller 404. The medium conveyance information includes a moving side address (From Address) and a moving destination address (To Address). The accessor controller 404 stores the medium conveyance information `XXXX` which was transferred from the director 402-1 into a conveyance information table 420 and instructs the conveying operation to the accessor 408-1 and, after that, supplies a response signal indicative of the reception to the director 402-1. In this instance, the execution flag is set into a slot of the accessor machine No. AS0 in the executing state management table 422. After confirmation of the reception response from the accessor controller 404, the director 402-1 reports a channel end status to the channel (B).
FIG. 3 is a diagram showing the operation in the case where a moving command of a logical machine No. address #0 was received from the Channel (A) subsequently to FIG. 2. In this instance, since the accessor 408-1 is conveying the medium by the command from the channel (B), when the moving command is generated from the channel (A), a "device busy" is reported. That is, when the moving command of the logical machine No. address #0 is received from the channel (A), the director 402-1 instructs a command reception to the accessor controller 404. The accessor controller 404 refers to the executing state management table 422. In this instance, since the execution flag of the accessor machine No. AS0 has been set, the accessor controller 404 sets a busy flag to the position of the director machine No. address DO in a busy management table 424 and returns a busy response to the director 402-1. The director 402 confirms the busy response from the accessor controller 404 and sets a busy report flag to the position of the channel (A) in the machine No. table 418-1 and reports a device busy status to the channel (A).
FIG. 4 shows the operation in the case where the moving command of the logical machine No. address #1 was received from the channel (E) subsequently to FIG. 3. In this case as well, since the accessor 408-1 is conveying the medium, a "busy" is finally reported to the channel (E).
FIG. 5 is a diagram showing the operation when the medium conveyance of the accessor 408-1 is finished. After confirmation of the end of the medium conveyance by the accessor 408-1, the accessor controller 404 reports the end to the director 402-1 and clears an execution flag of the executing state management table 422. After the conveyance end report was received from the accessor controller 404, the director 402-1 refers to the machine No. table 418-1 and reports a conveyance end status to the channel (B) to which the moving command was issued and clears the execution flag at the position of the channel (B).
FIG. 6 is a diagram showing the reporting operation of a busy cancellation after completion of the medium conveyance of FIG. 5. After confirming that the execution flag of the executing state management table 422 was turned off, the accessor controller 404 reports the busy cancellation to the director 402-1 and clears a busy flag of the director machine No. address D0 in the busy management table 424. After confirming the busy cancellation report from the accessor controller 404, the director 402-1 reports a busy cancellation status to the channel (A) in which a busy report flag was set. The director 402-1 further clears the busy report flag in the machine No. table 418-1. After the busy cancellation status from the director 402-1 was received, the channel (A) designates the logical machine No. address #0 and again generates a moving command.
FIG. 7 shows operating sequences in FIGS. 2 to 6 in the conventional apparatus in a lump. A moving command from the channel (B) is received by the director 402-1 and transferred as a medium conveying command 502 to the accessor controller 404. The accessor controller 404 returns a receipt response 504 and reports a zero status 506 to the channel (B). Subsequently, the director 402-1 receives medium conveyance information 508 from the channel (B) and transfers as conveyance information 510 to the accessor controller 404 and gives a conveying command 516 to the accessor 208-1. At the same time, the accessor controller 404 returns a receipt response 512. The director 402-1 reports a channel end status 514. A moving command 518 from the channel (A) is received by the director 402-1 and transferred as a medium conveying command 520 to the accessor controller 404. The accessor controller 404 returns a busy response 522. The director 402-1 reports a device busy status 524 to the channel (A). A moving command 526 from the channel (E) is received by the director 402-2 and transferred as a medium conveying command 528 to the accessor controller 404. The accessor controller 404 returns a busy response 530. The director 402-2 reports a device busy status 532 to the channel (E). It is now assumed that the medium conveyance by the accessor 408-1 had been finished and a conveyance end report 534 was sent to the accessor controller 404. A conveyance end report 536 is sent to the director 402-2. The director 402-2 reports a device end status 538 to the channel (B). The director 402-1 subsequently receives a busy cancellation 540 and reports a busy cancellation status 542 to the channel (A). In response to such a report, a moving command 544 is again generated from the channel (A).
According to the above conventional apparatus, the directors 402-1 and 402-2 have the machine No. tables 418-1 and 418-2 of an amount of only one address and the accessor controller 404 holds only one medium conveyance information. Therefore, the library apparatus can allocate only one logical machine No. address. The apparatus can execute only a successive process such that only after one moving command was completely finished for the moving command from the channel, the next moving command can be executed.
Therefore, in the case where the library apparatus has already been executing one moving command, the next moving command can be executed after the device end status indicative of the end of the command which is at present being executed was reported. Therefore, the conventional library apparatus has the following problems.
First, since the library apparatus cannot designate a plurality of machine numbers, a plurality of moving commands for the different machine numbers from the same channel cannot be executed.
Second, the library apparatus can accept only one of the moving commands generated simultaneously from a plurality of channels.
Third, during the execution of the moving command by a certain director, the library apparatus cannot accept the moving command by another director.
Fourth, in spite of the fact that the library apparatus has a plurality of accessors as a medium conveying mechanism, the number of accessors which execute the conveying operation is always set to one, so that a use efficiency is bad.
On the other hand, in the conventional library apparatus, there is a case where a placement error fault or an interface system fault occurs during the conveyance of the memory medium. FIG. 8 shows a situation of the occurrence of faults in the library apparatus. The medium conveyance information in association with the moving command from the host computer 20 is transferred to the accessor controller 404 through the director 402 of the library apparatus. The accessor controller 404 which received the medium conveyance information generates a command to the accessor 408-1, thereby conveying the medium, for example, from a moving side address (1) in the cell block 410 designated by the medium conveyance information to a recording and reproducing apparatus (MTU) 416 designated by a moving destination address (2). 0n the other hand, when the medium is left in the MTU 416 because of some causes and, in this state, the accessor 408-1 conveys the medium to the MTU 416 on the basis of the moving command which was newly generated, a placement error fault occurs. Namely, when the operator tries to load the medium conveyed by a robot hand 418 of the accessor 408-1 by pushing the medium into the MTU 416, the medium will collide with another medium which has already been loaded and will not move. Such a state is recognized as a placement error fault occurrence. A fault response is returned to the host computer 20. In the case where such a placement error fault occurs, the accessor controller 404 executes a recovering process according to a flowchart shown in FIG. 9. A conventional idea of the recovering process for such a placement error fault is such that it is first regarded that the accessor itself has an abnormality and an alignment error occurred, so that the operation of the accessor is again executed by setting the alignment error to a fault counter measure position. If the fault is not recovered even by the above operation, the accessor is alternated to the spare accessor and the recovering process is similarly executed. If the fault is not recovered even by executing the above process, it is regarded that there are faults in both of the accessor or the placement error fault cannot be recovered in accordance with the presence or absence of the medium on the moving destination side, thereby finishing the processes as a fault.
The case of FIG. 8 will now be practically explained as an example in step S1 in FIG. 9, the moving side address (1) and moving destination address (2) are respectively designated on the basis of the medium conveyance information by the moving command which has already been received, and a command to convey the medium is generated to the accessor 408-1. When a status response of the occurrence of the placement error fault is obtained in step S2, step S3 follows and a repositioning process by the same accessor 408-1 is executed. Namely, in step S4, a command in which the moving destination address (2) is held unchanged and the robot hand 418 of the accessor 408-1 is set to the moving destination address (2) is generated. When a placement error fault again occurs in step S5 by the repositioning process, it is regarded that there is an alignment error of the accessor 408-1. The processing routine advances to step S6. In step S6, a command to convey the medium which was held by the robot hand 418 of the accessor 408-1 and in which the placement error occurred to a predetermined shunt address (3) in the cell block 410 is generated. In step S7, an alternating process for alternating to the spare accessor 408-2 is subsequently executed. In step S8, a command in which the moving destination address (2) is held unchanged and the shunt cell address (3) was set to the moving side address is generated to the accessor 408-2 which alternated. When a status response of the placement error fault is again obtained by the conveyance by the spare accessor 408-2 which alternated, it is likewise regarded that there is an alignment error of the accessor 408-2, so that step S11 follows. After a command to convey the medium held by the robot hand 418 of the accessor 408-2 to the shunt cell address (3) was generated, a command to convey the medium remaining in the recording and reproducing apparatus 416 as a moving destination address (2) to another shunt cell address (4) is generated in step S12. When no medium exists in the moving destination address (4) in step S13, the processing routine advances to step S14. It is Judged that there are faults in both of the accessors 408-1 and 408-2, so that the processes are finished as a fault. When the medium exists in the moving destination address (2), it is Judged that the placement error fault could not he recovered, so that the processes are finished as a fault.
A flowchart of FIG. 10 shows a conventional recovering process for a fault of the interface system. An idea of the conventional recovering process for the interface system fault is such that in the case where when a fault occurs in an interface bus 600-1 between a host computer 20 and a director 402, the bus 600-1 is switched to an interface bus 600-2 for alternation. In this state, the accessor controller 404 judges whether the medium to be conveyed by the accessor 408-1 exists in the moving side address or the moving destination address. The following processes are executed in accordance with the result of the judgement.
I. When the medium exists in the moving side address, the command is again retried. PA1 II. When the medium exists in the moving destination address, the processes are finished as a normal state. PA1 III. When no medium exists in both of the moving side address and the moving destination address, it is determined that there is a possibility such that the medium was not erroneously picked up, so that the accessor operation to prevent a damage of the medium is inhibited and a fault which needs to call a system engineer is set.
Explanation will now be practically made with reference to FIG. 8 as an example. In step S1, an instruction of the medium conveyance in which the moving side address (1) and the moving destination address (2) were respectively designated on the basis of the medium conveyance information by the moving command which had already been received form the upper-order apparatus is generated from the accessor controller 404 to the accessor 408-1. In step S2, when the occurrence of a fault of the interface bus system is recognized by the notification from the director 402, the processing routine advances to step S3. In step S3, the interface bus 600-1 between the director 402 and the host computer 20 is changed to the interface bus 600-2 as an alternating path and the apparatus waits for establishment of the using state of the alternating path. In step S4, a command to convey the medium existing in the moving destination address to the shunt cell address (3) is generated to the same accessor 408-1. When it is judged in step S5 that the medium exists in the moving side address (1) by the execution of the command by the accessor 408-1, since the medium is not conveyed yet, step S6 follows and the same moving command as that in step S1 is again generated. When the medium doesn't exist in the moving side address (1) in step S5, a check is made in step S7 to see if the medium exists in the moving destination address (2) or not. When the medium exists in the moving destination address (2), it is judged that the medium has already been conveyed, so that the processes are finished as a normal state in step S8. When no medium exists in the moving destination address (2), it is judged that the medium was dropped on the way to the moving destination address (2), so that the operation of the accessor 408-1 is inhibited in step S9 and the processes are finished as a system engineer call fault.
However, the conventional recovering methods for the placement error fault and interface system fault as mentioned above have the following problems. First, in the conventional recovering method of the placement error fault shown in FIG. 9, the placement error fault is regarded as an alignment error which is caused by an abnormality of the accessor and the recovering process is executed. Therefore, there is a problem such that it takes a very long time of five to six minutes until the placement error fault is recognized. In addition, a self error due to the collision of the mediums occurs three times until a fault is finally judged as a placement error fault. That is, although the self error which occurs when the media first collide occurs due to the placement error fault, in case of the remaining self errors of two times, the medium collision occurs because of the recovering process of the placement error fault. Therefore, there is a problem such that the medium is damaged or the like by a shock which is applied by the collision of the media. Further, in the conventional recovering process of the interface system fault, in the case where the interface system fault overlaps the placement error fault, since the medium has already been conveyed to the shunt cell, no medium exists in both of the moving side address and the moving destination address. There is, consequently, a problem such that it is judged that the medium was dropped on the way to the moving destination address, so that the accessor operation is inhibited and the library apparatus is stopped as a CE call fault.